大学生研究素养相关因素：2015—2025年系统文献综述。
完整摘要 · 中文译文
本系统文献综述基于2015年1月1日至2025年12月31日间发表的50项研究，分析了与大学生研究素养相关的因素。遵循PRISMA指南，研究选自Scopus、Web of Science和Science Direct数据库。研究结果显示，研究素养是一种多维能力，通过三个相互关联的维度发挥作用——个人维度（63%）、教学维度（23%）和制度维度（14%），该比例基于35项具有可提取定量数据的研究计算得出。个人因素显示出最强的统计学关联，其中信息素养（d = 0.82）和研究自我效能感（d = 0.76）产生了最大的合并效应量。教学因素虽然代表性较低，但与之相关的效应量最大，尤其是图书馆与教师合作（d = 0.94）和结构化教学（d = 0.78）。制度因素，包括框架和技术资源，始终与个人及教学变量的较大估计值相关，这一模式与放大作用相符——但并非证明其存在。该综述识别出从传统信息素养向批判性人工智能素养的范式转变，2023年至2025年间与人工智能相关的研究增长了410%。文献计量分析显示，研究在地理上集中于欧洲和亚洲（50%），凸显了全球南方视角的代表性不足。方法学局限性包括相关性研究设计占主导（36%）以及纵向研究的缺乏。 占12%。该研究提出一个解释性框架，其中制度条件作为情境前因，塑造教学机会，进而通过研究自我效能感与个体研究能力相关联，并以学科、学术阶段和先前数字能力作为调节变量；由此推导出四个可检验的命题。由于证据基础主要为相关性研究，这些发现应被视为关联而非因果效应；它们为设计旨在促进当代高等教育情境中批判性思维、信息评估和伦理研究实践的教育策略提供了基于证据的基础。
完整摘要 · 英文原文
This systematic literature review analyses factors associated with research literacy among university students based on 50 studies published between 1 January 2015 and 31 December 2025. Following PRISMA guidelines, studies were selected from Scopus, Web of Science, and Science Direct databases. The findings reveal research literacy as a multidimensional competency operating through three interconnected dimensions - personal (63%), pedagogical (23%) and institutional (14%), computed over the 35 studies with extractable quantitative data -. Personal factors showed the strongest statistical associations, with information literacy (d = 0.82) and research self-efficacy (d = 0.76) yielding the largest pooled effect sizes. Pedagogical factors, though less represented, were associated with the largest effect sizes, particularly library-faculty collaboration (d = 0.94) and structured instruction (d = 0.78). Institutional factors, including frameworks and technological resources, were consistently associated with larger estimates for personal and pedagogical variables, a pattern compatible with - but not demonstrative of - an amplifying role. The review identified a paradigmatic shift from traditional information literacy toward critical artificial intelligence literacy, with a 410% growth in AI-related studies between 2023-2025. Bibliometric analysis revealed geographical concentration in Europe and Asia (50%), highlighting underrepresentation of Global South perspectives. Methodological limitations include predominance of correlational designs (36%) and scarcity of longitudinal studies (12%). The study proposes an explanatory framework in which institutional conditions operate as contextual antecedents that shape instructional opportunity, which in turn relates to individual research competence through research self-efficacy, with discipline, academic stage and prior digital competence acting as moderators; four testable propositions are derived from it. Because the evidence base is predominantly correlational, these findings should be read as associations rather than causal effects; they provide an evidence-informed basis for designing educational strategies intended to foster critical thinking, information evaluation, and ethical research practices in contemporary higher education contexts.